We read with interest the paper by Badouraki et al. entitled ‘Evaluation of pelvic ultrasonography in the diagnosis and differentiation of various forms of sexual precocity in girls’ published in the November 2008 edition of Ultrasound in Obstetrics and Gynecology1. The authors evaluated the performance of ultrasonographic measurement of uterine length, uterine volume, ovarian volume, the anteroposterior diameter at the fundus divided by the anteroposterior diameter at the cervix (F/C ratio), and the morphological appearance of the ovaries in discriminating girls with normal puberty, premature thelarche or premature pubarche from girls with central precocious puberty (CPP).
Although the most valuable parameter for differentiating girls with CPP is the measurement of luteinizing hormone (either the baseline value or after stimulation with gonadotropin-releasing hormone (GnRH) or GnRH analogs), it has been reported in a recent consensus statement that pelvic ultrasound imaging is considered helpful as an adjunct to GnRH stimulation in differentiating CPP from premature thelarche2. This consensus statement reported cut-off values for uterine length ranging from 3.4 to 4.0 cm, and between 1.0 and 3.0 cm3 for ovarian volume2. The wide range described for ovarian volumes is both concerning and frustrating. Serious attention is then required when reporting new cut-off values in order to obtain more reliable and precise values, and only then may we be able to enhance the performance of ultrasonography in the diagnosis of CPP. When reading the paper by Badouraki et al., we were disappointed by several inconsistencies regarding the assessment of ovarian volume in the diagnosis of CCP.
When comparing the described cut-off values for ovarian volume and the associated sensitivity and specificity with the mean and SDs for each age interval (0–6 years, 6–8 years and 8–10 years), it is obvious that the authors made some mistakes in their analysis. For example, for the age interval of 0–6 years they reported a cut-off value for ovarian volume of 3.04 cm3, above which the described sensitivity in identifying CPP was 100%. However, the mean value described for girls with CPP at this age interval was 2.62 cm3, so the sensitivity using the cut-off of 3.04 cm3 was probably lower than 50%. We therefore calculated the estimated sensitivities and specificities for uterine length and volume, ovarian volume and F/C based on the area under a standard normal curve using the reported means and SDs for each parameter (Table 1). The differences observed between the reported and estimated sensitivities for the ovarian volume cut-off values cannot be explained by sample characteristics.
|Parameter||Mean ± SD||Sensitivity (%)||Specificity (%)|
|Age 0–6 years*|
|Uterine length (cm)||4.44 ± 1.14||2.81 ± 0.43||3.46||85.70||80.23||97.20||93.32|
|Uterine volume (cm3)||4.78 ± 2.6||1.91 ± 0.47||2.62||100.00||79.67||91.70||99.95|
|Ovarian volume (cm3)||2.62 ± 1.00||1.26 ± 0.54||3.04||100.00||33.72||97.10||99.95|
|F/C ratio||1.33 ± 0.32||0.93 ± 0.09||0.95||71.40||88.10||86.10||58.71|
|Age 6–8 years*|
|Uterine length (cm)||4.16 ± 0.73||3.27 ± 0.35||3.74||88.20||71.57||94.70||90.99|
|Uterine volume (cm3)||4.82 ± 1.59||2.58 ± 0.88||3.48||82.40||79.95||89.50||84.61|
|Ovarian volume (cm3)||2.77 ± 0.97||1.71 ± 0.67||3.35||100.00||27.76||89.50||89.27|
|F/C ratio||1.21 ± 0.18||1.02 ± 0.08||1.05||82.40||81.33||84.20||64.80|
|Age 8–10 years†|
|Uterine length (cm)||5.0 ± 1.2||3.5 ± 0.7||4.06||76.90||78.23||81.80||78.81|
|Uterine volume (cm3)||10.0 ± 6.5||3.5 ± 0.7||4.35||84.60||80.51||81.80||88.69|
|Ovarian volume (cm3)||4.2 ± 2.2||1.8 ± 0.8||4.46||80.80||45.22||88.50||99.95|
|F/C ratio||1.4 ± 0.3||1.1 ± 0.1||1.16||76.90||78.81||69.20||72.57|
When evaluating the ovarian volume cut-off values and the corresponding receiver–operating characteristics curves (which are described in Figure 3 of the original paper), it seems that the described cut-off values may be able to provide high sensitivity values. However, the data points for the ovarian volumes of each group displayed in Figure 3 of the original paper cannot correspond to the reported means and SDs, as is clearly shown in our Figure 1. Therefore, either the described means or the cut-off values must be incorrect. Because the cut-off values are far above the values of 1.2 cm3 and 1.0 cm3 described previously3, 4, we suppose that the reported means and SDs are correct. So, using the area under a standard normal curve we estimated new cut-off values for ovarian volume that would provide a sensitivity of 80%. For the age interval 0–6 years, the cut-off value is 1.78 cm3 and specificity 83%; for 6–8 years, the cut-off value is 1.96 cm3 and specificity 64%; and for 8–10 years, the cut-off value is 2.69 cm3 and specificity 87%. These estimated values are closer to those described previously in the literature. We ask for the authors to revise their data, and also to analyze their data in relation to the anteroposterior diameter at the fundus alone, as some authors have reported that this is the best isolated ultrasonographic parameter5. We also believe that this is the measurement of the prepubertal uterus with the best reproducibility, as sometimes it is not very easy to determine the boundary between the cervix and the vagina when measuring length, or between the uterus and its ligaments when measuring width.